1 /*
2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
5 *
6 * Definitions for the IP router.
7 *
8 * Version: @(#)route.h 1.0.4 05/27/93
9 *
10 * Authors: Ross Biro
11 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12 * Fixes:
13 * Alan Cox : Reformatted. Added ip_rt_local()
14 * Alan Cox : Support for TCP parameters.
15 * Alexey Kuznetsov: Major changes for new routing code.
16 * Mike McLagan : Routing by source
17 * Robert Olsson : Added rt_cache statistics
18 *
19 * This program is free software; you can redistribute it and/or
20 * modify it under the terms of the GNU General Public License
21 * as published by the Free Software Foundation; either version
22 * 2 of the License, or (at your option) any later version.
23 */
24 #ifndef _ROUTE_H
25 #define _ROUTE_H
26
27 #include <net/dst.h>
28 #include <net/inetpeer.h>
29 #include <net/flow.h>
30 #include <net/inet_sock.h>
31 #include <net/ip_fib.h>
32 #include <net/l3mdev.h>
33 #include <linux/in_route.h>
34 #include <linux/rtnetlink.h>
35 #include <linux/rcupdate.h>
36 #include <linux/route.h>
37 #include <linux/ip.h>
38 #include <linux/cache.h>
39 #include <linux/security.h>
40
41 /* IPv4 datagram length is stored into 16bit field (tot_len) */
42 #define IP_MAX_MTU 0xFFFFU
43
44 #define RTO_ONLINK 0x01
45
46 #define RT_CONN_FLAGS(sk) (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
47 #define RT_CONN_FLAGS_TOS(sk,tos) (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
48
49 struct fib_nh;
50 struct fib_info;
51 struct uncached_list;
52 struct rtable {
53 struct dst_entry dst;
54
55 int rt_genid;
56 unsigned int rt_flags;
57 __u16 rt_type;
58 __u8 rt_is_input;
59 __u8 rt_uses_gateway;
60
61 int rt_iif;
62
63 /* Info on neighbour */
64 __be32 rt_gateway;
65
66 /* Miscellaneous cached information */
67 u32 rt_mtu_locked:1,
68 rt_pmtu:31;
69
70 u32 rt_table_id;
71
72 struct list_head rt_uncached;
73 struct uncached_list *rt_uncached_list;
74 };
75
rt_is_input_route(const struct rtable * rt)76 static inline bool rt_is_input_route(const struct rtable *rt)
77 {
78 return rt->rt_is_input != 0;
79 }
80
rt_is_output_route(const struct rtable * rt)81 static inline bool rt_is_output_route(const struct rtable *rt)
82 {
83 return rt->rt_is_input == 0;
84 }
85
rt_nexthop(const struct rtable * rt,__be32 daddr)86 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
87 {
88 if (rt->rt_gateway)
89 return rt->rt_gateway;
90 return daddr;
91 }
92
93 struct ip_rt_acct {
94 __u32 o_bytes;
95 __u32 o_packets;
96 __u32 i_bytes;
97 __u32 i_packets;
98 };
99
100 struct rt_cache_stat {
101 unsigned int in_slow_tot;
102 unsigned int in_slow_mc;
103 unsigned int in_no_route;
104 unsigned int in_brd;
105 unsigned int in_martian_dst;
106 unsigned int in_martian_src;
107 unsigned int out_slow_tot;
108 unsigned int out_slow_mc;
109 };
110
111 extern struct ip_rt_acct __percpu *ip_rt_acct;
112
113 struct in_device;
114
115 int ip_rt_init(void);
116 void rt_cache_flush(struct net *net);
117 void rt_flush_dev(struct net_device *dev);
118 struct rtable *__ip_route_output_key_hash(struct net *, struct flowi4 *flp,
119 int mp_hash);
120
__ip_route_output_key(struct net * net,struct flowi4 * flp)121 static inline struct rtable *__ip_route_output_key(struct net *net,
122 struct flowi4 *flp)
123 {
124 return __ip_route_output_key_hash(net, flp, -1);
125 }
126
127 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
128 const struct sock *sk);
129 struct dst_entry *ipv4_blackhole_route(struct net *net,
130 struct dst_entry *dst_orig);
131
ip_route_output_key(struct net * net,struct flowi4 * flp)132 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
133 {
134 return ip_route_output_flow(net, flp, NULL);
135 }
136
ip_route_output(struct net * net,__be32 daddr,__be32 saddr,u8 tos,int oif)137 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
138 __be32 saddr, u8 tos, int oif)
139 {
140 struct flowi4 fl4 = {
141 .flowi4_oif = oif,
142 .flowi4_tos = tos,
143 .daddr = daddr,
144 .saddr = saddr,
145 };
146 return ip_route_output_key(net, &fl4);
147 }
148
ip_route_output_ports(struct net * net,struct flowi4 * fl4,struct sock * sk,__be32 daddr,__be32 saddr,__be16 dport,__be16 sport,__u8 proto,__u8 tos,int oif)149 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
150 struct sock *sk,
151 __be32 daddr, __be32 saddr,
152 __be16 dport, __be16 sport,
153 __u8 proto, __u8 tos, int oif)
154 {
155 flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
156 RT_SCOPE_UNIVERSE, proto,
157 sk ? inet_sk_flowi_flags(sk) : 0,
158 daddr, saddr, dport, sport, sock_net_uid(net, sk));
159 if (sk)
160 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
161 return ip_route_output_flow(net, fl4, sk);
162 }
163
ip_route_output_gre(struct net * net,struct flowi4 * fl4,__be32 daddr,__be32 saddr,__be32 gre_key,__u8 tos,int oif)164 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
165 __be32 daddr, __be32 saddr,
166 __be32 gre_key, __u8 tos, int oif)
167 {
168 memset(fl4, 0, sizeof(*fl4));
169 fl4->flowi4_oif = oif;
170 fl4->daddr = daddr;
171 fl4->saddr = saddr;
172 fl4->flowi4_tos = tos;
173 fl4->flowi4_proto = IPPROTO_GRE;
174 fl4->fl4_gre_key = gre_key;
175 return ip_route_output_key(net, fl4);
176 }
177
178 int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
179 u8 tos, struct net_device *devin);
180
ip_route_input(struct sk_buff * skb,__be32 dst,__be32 src,u8 tos,struct net_device * devin)181 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
182 u8 tos, struct net_device *devin)
183 {
184 int err;
185
186 rcu_read_lock();
187 err = ip_route_input_noref(skb, dst, src, tos, devin);
188 if (!err)
189 skb_dst_force(skb);
190 rcu_read_unlock();
191
192 return err;
193 }
194
195 void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
196 u32 mark, u8 protocol, int flow_flags);
197 void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
198 void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
199 u8 protocol, int flow_flags);
200 void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
201 void ip_rt_send_redirect(struct sk_buff *skb);
202
203 unsigned int inet_addr_type(struct net *net, __be32 addr);
204 unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
205 unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
206 __be32 addr);
207 unsigned int inet_addr_type_dev_table(struct net *net,
208 const struct net_device *dev,
209 __be32 addr);
210 void ip_rt_multicast_event(struct in_device *);
211 int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
212 void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
213 struct rtable *rt_dst_alloc(struct net_device *dev,
214 unsigned int flags, u16 type,
215 bool nopolicy, bool noxfrm, bool will_cache);
216
217 struct in_ifaddr;
218 void fib_add_ifaddr(struct in_ifaddr *);
219 void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
220
ip_rt_put(struct rtable * rt)221 static inline void ip_rt_put(struct rtable *rt)
222 {
223 /* dst_release() accepts a NULL parameter.
224 * We rely on dst being first structure in struct rtable
225 */
226 BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
227 dst_release(&rt->dst);
228 }
229
230 #define IPTOS_RT_MASK (IPTOS_TOS_MASK & ~3)
231
232 extern const __u8 ip_tos2prio[16];
233
rt_tos2priority(u8 tos)234 static inline char rt_tos2priority(u8 tos)
235 {
236 return ip_tos2prio[IPTOS_TOS(tos)>>1];
237 }
238
239 /* ip_route_connect() and ip_route_newports() work in tandem whilst
240 * binding a socket for a new outgoing connection.
241 *
242 * In order to use IPSEC properly, we must, in the end, have a
243 * route that was looked up using all available keys including source
244 * and destination ports.
245 *
246 * However, if a source port needs to be allocated (the user specified
247 * a wildcard source port) we need to obtain addressing information
248 * in order to perform that allocation.
249 *
250 * So ip_route_connect() looks up a route using wildcarded source and
251 * destination ports in the key, simply so that we can get a pair of
252 * addresses to use for port allocation.
253 *
254 * Later, once the ports are allocated, ip_route_newports() will make
255 * another route lookup if needed to make sure we catch any IPSEC
256 * rules keyed on the port information.
257 *
258 * The callers allocate the flow key on their stack, and must pass in
259 * the same flowi4 object to both the ip_route_connect() and the
260 * ip_route_newports() calls.
261 */
262
ip_route_connect_init(struct flowi4 * fl4,__be32 dst,__be32 src,u32 tos,int oif,u8 protocol,__be16 sport,__be16 dport,struct sock * sk)263 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
264 u32 tos, int oif, u8 protocol,
265 __be16 sport, __be16 dport,
266 struct sock *sk)
267 {
268 __u8 flow_flags = 0;
269
270 if (inet_sk(sk)->transparent)
271 flow_flags |= FLOWI_FLAG_ANYSRC;
272
273 flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
274 protocol, flow_flags, dst, src, dport, sport,
275 sk->sk_uid);
276 }
277
ip_route_connect(struct flowi4 * fl4,__be32 dst,__be32 src,u32 tos,int oif,u8 protocol,__be16 sport,__be16 dport,struct sock * sk)278 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
279 __be32 dst, __be32 src, u32 tos,
280 int oif, u8 protocol,
281 __be16 sport, __be16 dport,
282 struct sock *sk)
283 {
284 struct net *net = sock_net(sk);
285 struct rtable *rt;
286
287 ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
288 sport, dport, sk);
289
290 if (!src && oif) {
291 int rc;
292
293 rc = l3mdev_get_saddr(net, oif, fl4);
294 if (rc < 0)
295 return ERR_PTR(rc);
296
297 src = fl4->saddr;
298 }
299 if (!dst || !src) {
300 rt = __ip_route_output_key(net, fl4);
301 if (IS_ERR(rt))
302 return rt;
303 ip_rt_put(rt);
304 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
305 }
306 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
307 return ip_route_output_flow(net, fl4, sk);
308 }
309
ip_route_newports(struct flowi4 * fl4,struct rtable * rt,__be16 orig_sport,__be16 orig_dport,__be16 sport,__be16 dport,struct sock * sk)310 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
311 __be16 orig_sport, __be16 orig_dport,
312 __be16 sport, __be16 dport,
313 struct sock *sk)
314 {
315 if (sport != orig_sport || dport != orig_dport) {
316 fl4->fl4_dport = dport;
317 fl4->fl4_sport = sport;
318 ip_rt_put(rt);
319 flowi4_update_output(fl4, sk->sk_bound_dev_if,
320 RT_CONN_FLAGS(sk), fl4->daddr,
321 fl4->saddr);
322 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
323 return ip_route_output_flow(sock_net(sk), fl4, sk);
324 }
325 return rt;
326 }
327
inet_iif(const struct sk_buff * skb)328 static inline int inet_iif(const struct sk_buff *skb)
329 {
330 int iif = skb_rtable(skb)->rt_iif;
331
332 if (iif)
333 return iif;
334 return skb->skb_iif;
335 }
336
337 extern int sysctl_ip_default_ttl;
338
ip4_dst_hoplimit(const struct dst_entry * dst)339 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
340 {
341 int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
342
343 if (hoplimit == 0)
344 hoplimit = sysctl_ip_default_ttl;
345 return hoplimit;
346 }
347
348 #endif /* _ROUTE_H */
349